NVIDIA's AVO harness lifts Opus 5 from 30% to 100% on ARC-AGI-3
daniel_mac8 · x · 2026-08-22
NVIDIA solved the ARC-AGI-3 public benchmark with its agent harness AVO + Opus 5, boosting Opus 5 from 30% under the standard harness to 100%. The poster breaks down what the harness does to expose an LLM's latent capabilities:
- Persistent memory across context windows
- An inspect → plan → implement → evaluate loop
- Execution feedback so the agent learns from mistakes at inference time
- A supervisor that detects stagnation in failed trajectories and redirects
- External statefulness of the transcript (tool calls, execution artifacts, agent trajectory)
His takeaway: frontier harness engineering doesn't require working at a frontier lab — harness advances don't need to touch model weights but can make all the difference.
Related event: NVIDIA's AVO Coding Agent Scores 100% on ARC-AGI-3(12 posts)→
More from coding & agent
- Agent coding hack: Multi-instance collaboration to optimize UI/UX — doodlestein · 2026-08-22
- xAI Launches Grok Bot: AI Teammates That Log Into Your Tools and Do Real Work — tetsuoai · 2026-08-22
- Developer criticizes closed-source models for hiding Chain of Thought — johnnyApplePRNG · 2026-08-22
- An Agent is 3 Layers: Business Logic, Harness, and Infrastructure — blaizedsouza · 2026-08-22
- Every Agent gains computer control, experiments reveal high token costs — every · 2026-08-22
- Grok Build 1.0.8 ships: concurrent subagents start faster, no more frozen sessions — kevinnbass · 2026-08-22